視的長さの繼時比較に於ける時間誤差

抄録

In another paper (1), I have pointed out that the negative time-error was dominant in the suc cessive comparison of visual lengths. But American experimenters report recently that the positive time-error may arise, too (2 etc). The purpose of this experiment is to make sure of this situation and to determine from what factors these differences result.
Lines of various lengths were presented to the subjects by projection on a sdreen. The standard line was 200mm long and nine comparison stimuli were 240, 230, 220, 210, 200, 190, 180, 170, and 160 mm. in length. After a signal the first stimulus was presented for 1.5 sec. and after a 1.5, 3, 4. 5, 6, 8 and 12 sec. interval the second stimulus was presented also for 1.5 sec. The subjects were asked to write down immediately whether the second line appeared longer than, equal to, or shorter than the first line.
In these experiments, I could get the following results.
1. In all the cases, the time-error was negative, and no positive results were obtained.
2. The difference between conditions of the pre-presentation of the stimulus and its post-presentation had no significantly different effect on the time-error. (In this case only a 8 sec. interval was used.)
3. The enlargement of the experimental series reliably shifts the time-errors in the positive direction and eliminates the negative time-error altogether. (In this case the conditions were the as those in 1.)
In the other experiment in which the dominant personality group, and the submissive personality group, rated by a personality inventory, were asked to draw the medium live out of one-serial lines, we could not find any significant correlation between the personality traits and the drawn lines. But when this experiment was repeated many times, we found shifts in judgement scale.
Moreover, it was found that the way the time-error occurred is different, depending on the children above or below the 4th grade in the primary school. There were found reliable differences in the results between normal and abnomal children. The enlargement of the experimental series uniformly and reliably shifted the time-errors in the possive direction in normal children, but the results failed to show this tendency for abnormal children.